Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z...
Transcript of Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z...
![Page 1: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/1.jpg)
Office of Hydrologic Development
Silver Spring, MD
National Oceanic and Atmospheric Administration’s
National Weather Service
Hydrologic Ensemble Forecasting Service
(HEFS)
Seminar H
HEFS ConOps Discussion
Mark Fresch
HEFS Training Workshop 5, Silver Spring, MD
February 25-26, 2013
![Page 2: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/2.jpg)
Office of Hydrologic Development
Silver Spring, MD
National Oceanic and Atmospheric Administration’s
National Weather Service
HEFS Components
Page: H-2
![Page 3: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/3.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Objective
Objective: To get feedback from HEFS RFCs on
how to run HEFS in routine CHPS operations
o Project planning for the rollout will be discussed during
Seminar I – Project Status and Plans
o Feedback (seminars H and I) will be included in the HEFS
ConOps Document, to guide the rollout
o ConOps has recommendations and options
o OHD and HEFS test RFCs will iterate on the document
through mid-April
o OHD and all RFCs will finalize the document by the end of
May
Seminar I Page: H-3
![Page 4: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/4.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Outline
Outline o Implementation
o Operational Runs
o MEFP
o MODs
o Calibration - Parameter Estimation
o Hindcasting and Verification
o Archiving
o New Products
o Training and Out-reach
Seminar I Page: H-4
![Page 5: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/5.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Implementation
On average, how much time has your RFC spent per week on HEFS?
o 20 hours per week (last six months)
o 50% of one person
o CN & NE 10-20 hours
What part of implementation could be improved?
o By OHD (training, software, documentation, releases)
o By you/RFC (more efficient implementation, participation in team)
o Strategy of incremental rollout?
• All points or by Forecast group
• Records of data available
• Users in mind
• Hardware issues & configuration management
• Keep consistent with operational configuration
o Expectations of hindcasting/verification?
• Not a crazy idea
Seminar I Page: H-5
![Page 6: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/6.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Implementation
Rank the priorities of HEFS software/science tasks (High, Med, Low)
Seminar I Page: H-6
Task Priority
MEFP – fix probability of precipitation bias
MEFP – evaluating and adjust for extreme events
MEFP – use grids after 00Z (for 12Z TØ)
MEFP – running at TØ other than 12Z
MEFP – canonical events/Schaake Shuffle concerns (FogBugz 979)
MEFP – add other forcings sources, e.g. WPC QPF
MEFP – large day-to-day changes to long range forecasts
Other MEFP issues?
![Page 7: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/7.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Implementation
Rank the priorities of HEFS software/science tasks (High, Med, Low)
Seminar I Page: H-7
Task Priority
EnsPost – run at 6 hr time steps
EnsPost – evaluate and improve at regulated points
EnsPost – enhance to use on ephemeral streams
Add Data Assimilator (account for uncertainty in initial conditions)
Run-time issues (MEFP, EnsPost, Parameter Estimators, HEFS,
EVS, hindcasting, or ensemble runs in general)1
Usability issues (MEFP, EnsPost, Parameter Estimators, HEFS,
EVS, hindcasting, ensemble runs in general)
Others?
![Page 8: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/8.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Implementation – Coverage
After two years:
ABRFC: MEFP at 440 basins for precip. & 103 basins for temp.; streamflow at 239
pts; and EnsPost and GraphGen at ~140 of those pts
CBRFC: MEFP at 317 basins & streamflow for ~240 pts and adding EnsPost
CNRFC: MEFP at 319 basins; streamflow at 199 pts.; and EnsPost at 30 pts. Plans to
expand EnsPost & add GraphGen
MARFC: MEFP at ~100 basins and streamflow and GraphGen at 53 pts (the Del. R.)
for internal use and a second run of MEFP and streamflow for (14) NYCDEP points
NERFC: MEFP at 12 basins; streamflow and EnsPost at 6 pts for internal use and a
second run of MEFP and streamflow for (8) NYCDEP points
Seminar I Page: H-8
![Page 9: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/9.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Operational Runs
What are the advantages and recommendations?
o Automatic v. interactive
• What is being done interactively at NE and CN?
– Manual run (NE & CN)
• Is it reasonable to do HEFS forecasts interactively (via IFD)
one point and step at a time?
o Configuring workflows – advantages to different scenarios
• Separate runs for MEFP, precip and temp?
• Separate runs for different HEFS components?
• Separate runs for small number of points?
– Up to each RFC’s operational needs
Seminar I Page: H-9
![Page 10: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/10.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Operational Runs
Timing of runs (in red)
• Grid downloads
• CFSv2: 4x daily {0Z,6Z,12Z,18Z} with 28hr lag to download (12Z/D1 16Z/D2)
• GEFS: 1x daily {0Z} with 8hr lag to download (i.e. 8Z)
• Grid ingest into CHPS
• CFSv2: 1-hour lag from download (5Z, 11Z, 17Z, & 23Z)
• GEFS: 1-hour lag from download (9Z)
• MEFP ensemble generation
• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not
using the latest (or any) CFSV2
• Streamflow ensemble generation
• 1x daily with T0=12Z, but executed at 17:30Z
• Same time as MEFP ensemble generation
• By Forecast Group (i.e. matches ESP)
Seminar I Page: H-10
![Page 11: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/11.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Operational Runs
Forecast horizon(s)
o One long range run for all RFCs needs or separate runs for
different forecast horizons?
o If separate runs
• Short-medium range: daily
• Long range: to meet RFC needs, ala ESP runs
• CN – 14 day (10 mins) and 365 day
• More runs during flooding
Seminar I Page: H-11
![Page 12: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/12.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
MEFP
Forcings – available forecast horizon: recommendation
o RFC QPF (0 to X days): recommended if ‘large’ archive
available
o RFC QTF (0 to X days) – recommended, but believe
there’s no supportive archive
o GEFS (0 to 15 days) – recommended
o CFSV2 (0 to 270 days) – optional; rarely better the
climatology; RFCs should validate for their domain
o Climatology
• MEFP raw climatology or CHPS raw climatology - recommended
(for operational runs)
• MEFP resampled climatology – not recommended
Seminar I Page: H-12
![Page 13: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/13.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
MODs
Manual changes to ensemble traces and means – not recommended
MODs – define modifiersGroup
o Okay to use; others not used with ensembles (e.g. CN) <modifiersGroup id="ensembles">
<modifierId>tschng</modifierId>
<modifierId>setqmean</modifierId>
<modifierId>chgblend</modifierId>
<modifierId>setmsng</modifierId>
<modifierId>rrichng</modifierId>
<modifierId>mfc</modifierId>
<modifierId>uadj</modifierId>
<modifierId>switchTs</modifierId>
<modifierId>uhgchg</modifierId>
</modifiersGroup>
MODs used in ESP?
MODs may conflict with EnsPost – do some validation
Seminar I Page: H-13
![Page 14: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/14.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Calibration – Parameter Estimation
Calibration - # of years of archive needed1
o Less than 5 years, is not recommended - problematic for MEFP or
EnsPost.
o At least 10 years is highly recommended
o 25 years should provide a solid calibration
Frequency of calibration - consistency v. operational use
o Changes to CHPS configuration
o Recalibration of or changes to hydro. models
• Changes to ratings or hydrology
o New or updated forcings reforecast, e.g. driven by a major updates to
models
o Some changes to HEFS – in release notes; we are trying to minimize
these
o Anticipated frequency ~ 1-2 years; is this reasonable?
Seminar I Page: H-14
![Page 15: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/15.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Hindcasting and Verification
Recommend hindcasting/verification at RFCs in
partnership with OHD
o Provide objective guidance for better implementation
o Validate HEFS as the source of public products, such as
AHPS - OHD/RFCs need to develop criteria and plans
o Practical tips - already provided & in hindcasting guide
Seminar I Page: H-15
![Page 16: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/16.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Archiving
What to archive – recommended at RFCs
o Latest 30 days of CFSV2 and GEFS grids (move to NWC?)
o RFC QPF (already archived?)
o MEFP temperature and precip ensembles
o Streamflow ensembles based on MEFP with climatological
forcing (if produced)
o Streamflow ensembles based on your operational ESP
o Streamflow ensembles without and with (if produced)
EnsPost applied
Format – recommended (smallest output) for EVS
o .fi/.bin – fastinfoset / binary
Seminar I Page: H-16
![Page 17: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/17.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service Seminar I Page: H-17
New product: 10-day streamflow forecast with
uncertainty bounds (testing at ABRFC)
Provide streamflow forecasts as data (time
series) ? If so, to what forecast horizon(s)?
Others?
Any new products should have a baseline GG
template – agreed? -
o Yes
(Public) Products
![Page 18: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/18.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Training and Out-reach
At your office, besides HEFS focal points, who (other RFC
staff, WFOs, stakeholders (e.g. emergency managers)
need training on:
o HEFS functionality and mechanics (using/implementing HEFS)?
o Hydro. ensembles and probabilities (understanding HEFS forecasts)?
Assuming they need training, how / who does the
training? RFCs, OHD, on-line training/documents?
o RFC Staff – DOH and HEFS Focal Point?
o WFOs
o Stakeholders (emergency managers)
o Others
Seminar I Page: H-18
![Page 19: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/19.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Training needs (beyond HEFS focal points)
Group HEFS
use/impl.
Trainer Hydro.
Ens/Prob
Trainer
Other RFC staff Everyone HEFS Focal
point
WFOs Yes RFC/OHD?
Stakeholders
Seminar I
![Page 20: Hydrologic Ensemble Forecasting Service (HEFS)...• 1x daily with T0=12Z, but executed at 17:30Z (shifted back); or as early as 9Z, if not using the latest (or any) CFSV2 • Streamflow](https://reader033.fdocuments.net/reader033/viewer/2022053010/5f0decae7e708231d43cc142/html5/thumbnails/20.jpg)
National Oceanic and Atmospheric Administration’s
National Weather Service
Questions and comments?
Seminar I Page: H-20